Shulou(Shulou.com)05/31 Report--

In view of what problems need to be solved in Oracle practice, this article introduces the corresponding analysis and solutions in detail, hoping to help more partners who want to solve this problem to find a more simple and feasible way.

"de-Oracle" has been one of the most common words in describing system architecture transformation in the last 10 years. Although "de-Oracle" is talked about by many engineers and technical practitioners, there are not many cases in which the industry can really Oracle all the systems, especially the core systems of the financial scenario. So what's the difficulty of going to Oracle?

To answer this question, we first need to understand the nature of de-Oracle architecture transformation. One of the essence of the transformation of de-Oracle architecture is to make the system architecture have the ability to replace databases online, whether the target database of Oracle is MySQL or other relational databases, it is ultimately to solve such a problem.

What is the difficulty of replacing the database online and what problems will be encountered?

1. How to migrate real-time dynamic data without perception?

First of all, the database is not one of the core components of the transactional system, which is no exaggeration in the evaluation of the importance of the database. At present, most well-known websites and systems provide 7x24 services 24 hours a day, and the database is also dealing with a large number of read and write services 24 hours a day. To achieve Oracle means that you have to let a MySQL library quickly replace the Oracle library at some point while a MySQL library is still providing services, and smoothly take over all the services of Oracle.

Different from the stateless system component switching, which cuts off the traffic and completes the switching work, as a stateful system component, how to design a set of safe de-Oracle scheme from application transformation, to data synchronization, and then to database traffic switching, we can be very careful to migrate the data on the Oracle library which is providing services and the data is in a real-time changing state to MySQL seamlessly.

two。 How to manage and coordinate the high-frequency iterative de-Oracle transformation and functional transformation?

Secondly, the main work of Oracle architecture transformation is the refactoring of the application layer code, especially for the DAO (data access layer) refactoring. For some complex systems, the refactoring time will last several months or even longer. In this long time window of Oracle transformation, not only the data of Oracle library is changing dynamically, but also the functional code of the application is constantly changing for a website and system in high-speed iteration. If team An is doing Oracle architecture transformation for the application, and team B is constantly developing new functions for the application during this period, how to split the de-Oracle work can effectively manage and coordinate the coding and online rhythm of the two development teams.

3. How to safely switch the flow of landing database online?

When the Oracle transformation of the application of a library is completed and launched, the traffic of the production environment Oracle will be switched to MySQL. In this switching process, how to ensure that the last transaction on Oracle is successfully committed, and complete the data consistency check after synchronizing to MySQL, and all writes to this Oracle library can be quickly switched to MySQL without the abnormal state of partial write traffic Oracle, partial write traffic MySQL, and double write of the two libraries.

When the traffic is switched to MySQL, if the application reports an error or the performance problems of bug or MySQL are not covered in the previous stress test or preparation, how to quickly fail back the traffic to the Oracle and ensure that the data written in the MySQL can be returned to the Oracle exactly the same.

4. How to effectively split the task to Oracle?

In order to realize the de-Oracle of the whole station, it is necessary to transform some complex and huge core systems to de-Oracle. Take lufax as an example, the subsystems that provide basic services to all financial product lines in the whole station, such as user centers, asset centers, and capital accounts, are such complex and huge core systems. at the same time, including funds, main accounts and other exclusive financial product lines of business logic is complex, so the subsystem is also very large.

Therefore, for such subsystems, if you need to complete the transformation of all the Oracle in a large version and switch all from Oracle to M at one time during the business trough in one night, whether it is the switching risk of that night, or after the switching is completed, the risk of problems and bug in the peak period of the next day, including the workload of the development team to Oracle transformation in a short time and the probability of major bug are very large and uncontrollable.

How to split a large and important complex subsystem into multiple de-Oracle versions and switch traffic according to batches and make the impact of a single batch controllable is also a scheme that needs to be carefully designed in the whole station to Oracle.

The above mentioned four major problems that need to be solved to implement online library replacement at the architecture layer in Oracle. In addition to online database exchange, the second essence of de-Oracle architecture transformation is to introduce more storage engines in appropriate scenarios to undertake the computing and storage capacity of Oracle databases. This leads to the fifth question.

5. How to use an appropriate open source storage engine to Oracle and integrate architectures

First of all, Oracle is a very powerful relational database, which performs well in both OLTP and OLAP scenarios, and has a set of complete and easy-to-use operation and monitoring tools. But at the same time, although Oracle supports a variety of scenarios, some open source databases or storage engines outperform Oracle in terms of performance or cost investment in specific scenarios, which is a more appropriate choice than Oracle.

Therefore, going to Oracle is not only about going to Oracle to MySQL, MySQL can only undertake some of the computing and storage capacity of Oracle. During the landing process of lufax's Oracle, in addition to MySQL, we also use storage engines such as ES, HBase, TiDB, Impala+kudu and even application layer code to undertake and replace Oracle in different scenarios, and the overall revenue is better than using Oracle.

The above are the five major categories of actual combat problems encountered by lufax in the process of going to Oracle. There are still many details that need to be solved in the process of going to O, which cannot be enumerated here, because going to Oracle as a complex system architecture transformation itself requires the technical team to deal with all kinds of details.

The answers to the problems that need to be solved in Oracle practice are shared here. I hope the above content can be of some help to you. If you still have a lot of doubts to be solved, you can follow the industry information channel for more related knowledge.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.